I analyzed user feedback at Collision Conf. using Gemini Multimodal Exprimental Model (Vertex AI).
The annual Collision Conference brings together a vibrant community of entrepreneurs, tech enthusiasts, and industry leaders. This year, I attended Collision Conf with a specific goal: to leverage the power of AI to gain deeper insights from user feedback.
The traditional approach to conference feedback often involves surveys and paper forms. However, these methods can be time-consuming to analyze and may not capture the full range of participant opinions. This is where the Gemini Multimodal Experimental Model from Google's Vertex AI comes in.
The Gemini Multimodal Experimental Model offers a groundbreaking approach to analyzing user feedback, particularly when presented visually. Using a technique called prompt engineering, I was able to "teach" Gemini to interpret user responses captured on posters, whiteboards, and even world maps – data formats that would be challenging to analyze manually.
Following are my observations and findings:
Automation
Original Image
Model Output
Text
Title:
Automation Survey
Description:
The image shows a black poster board with white text and yellow circles representing data points. The poster is titled "COLLISION" and contains three survey questions related to automation. The first question asks about industry potential for automation with three options, each represented by a circle filled with yellow dots. The second question asks about perceived benefits of automation with four options, each represented by yellow dots in rectangular shapes. The third question explores factors holding back greater adoption of automation with four options represented by yellow dots in rectangular shapes.
Datapoints:
Question 1: "In which industry do you see the most potential for automation?"
• Transportation: Has a moderate amount of yellow dots representing responses.
• Distribution and Logistics: Has the highest amount of yellow dots, indicating the highest potential for automation according to the survey.
• Construction and Infrastructure: Has a lower amount of yellow dots compared to the other two options.Question 2: "What do you see as the benefits of automation?"
• Improved health and safety: Has a moderate amount of yellow dots.
• Cost-saving: Has a moderate amount of yellow dots.
• Efficiency of service: Has the highest amount of yellow dots among the benefits options.
• Increased productivity: Has a high amount of yellow dots, suggesting it's a significant perceived benefit of automation.Question 3: "What factors are holding back greater adoption of automation?"
• Labour relations: Has a high amount of yellow dots, suggesting it's a significant factor holding back automation adoption.
• The pace of technological advancement: Has the lowest amount of yellow dots, indicating it's not a major concern.
• Cost: Has a moderate amount of yellow dots.
• Government regulation: Has a moderate amount of yellow dots.Insights: Based on the data visualization, the survey suggests that distribution and logistics are perceived as having the highest potential for automation. The most common perceived benefits of automation are efficiency of service and increased productivity. However, labour relations seem to be the biggest hurdle to greater adoption of automation.
In my opinion...
Distribution and logistics emerged as the industry with the most perceived potential for automation. Interestingly, efficiency of service and increased productivity were considered the top potential benefits. However, a significant portion of respondents identified labor relations as a major hurdle to wider adoption. This highlights the need for open communication and collaboration between businesses and their workforce as automation continues to evolve.
AI is revolutionizing the supply chain. As observed by McKinsey, it’s boosting efficiency with smarter logistics, optimized inventory, and predictive maintenance. It also unlocks sharper demand forecasting and proactive risk management, all while providing real-time visibility. Depending on labor displacement, implementation costs, and data concerns require careful consideration as well.
AI
Original Image
Model Output
Text
Title:
AI Perception Survey
Description:
The image shows a black board with white text and yellow dot stickers. The board is divided into sections, each asking a question about artificial intelligence (AI) and its ethical principles, potential applications, and public perception. The questions are answered by respondents using the yellow dot stickers to mark their opinions on a scale or within different categories.
Data Points and Insights:
Ethical Principles of AI:
• The image includes a horizontal line representing a spectrum of opinions on the importance of ethical principles in AI development and use.
• The yellow dot stickers show that a majority of respondents believe ethical principles are very important, with a smaller number of dots clustered around the "neutral" point and very few dots near the "not important" end.AI Applications:
• The board presents four categories where AI could potentially have the most positive impact: housing and affordability, healthcare, education, and economy/environment/climate.
• The number of yellow dots in each category indicates respondent opinion on the potential benefits of AI in each area. The highest concentration of dots is in healthcare, followed by education, then housing and affordability, and lastly economy/environment/climate.Public Trust in AI:
• A scatter plot asks whether respondents feel informed enough about AI to trust its implications.
• The dots are distributed across a grid with "Yes" and "No" options on both the x and y axis.
• The distribution suggests that a majority of respondents fall into the "yes" quadrant, indicating that they feel reasonably informed about AI and its implications.AI and Existing Biases:
• A scatter plot explores whether respondents believe AI amplifies existing biases in society.
• The dot distribution indicates a mixture of opinions. A significant portion of respondents believe AI does amplify existing biases, but there are also a considerable number who believe it does not.
In my opinion...
A strong majority of respondents believed ethical principles were crucial in AI development and use. This illustrates the growing public consciousness about responsible AI development. The survey also identified healthcare as the area with the most potential for positive AI impact, followed by education and housing affordability.
As observed by McKinsey, AI is transforming healthcare, boosting diagnostic accuracy, personalizing treatment plans, and accelerating research. Imagine leveraging AI for early disease detection, robotic-assisted surgery with improved precision, or 24/7 virtual patient support. These are just a few ways AI is empowering healthcare professionals. However, as we embrace AI, ethical considerations are critical. Biased algorithms can perpetuate disparities in care. Robust data security and patient privacy are paramount. AI should augment, not replace, human expertise. Transparency in AI decision-making builds trust and ensures responsible implementation.
Cryptocurrency
Original Image
Model Output
Text
Title:
Cryptocurrency Panel
Description:
The image shows a black panel with white text and yellow circular graphics. The panel is likely from a conference or event and contains questions related to cryptocurrency, including the viability of Bitcoin as an alternative to fiat currency, the impact of real-world asset tokenization on real estate and commodities markets, and public understanding of staking in cryptocurrency. The questions are accompanied by visual representations using yellow circles, likely representing responses to a survey or poll.
Datapoints and Insights:
1. Bitcoin Viability:
The image contains a horizontal line graph representing the viability of Bitcoin, with "Not viable at all", "Somewhat viable", "Moderately viable", and "Very viable" marked along the line.
The graph shows a cluster of yellow circles towards the "Moderately viable" area, suggesting that the majority of respondents perceive Bitcoin as moderately viable as an alternative to fiat currency.2. Real-World Asset Tokenization:
The panel includes four boxes representing potential impacts of real-world asset tokenization, with labels "Increased market efficiency", "Reduced transaction costs", "Enabled access for a new generation of investors", and "It depends on regulatory adaptations".
Each box contains a scatterplot of yellow circles, suggesting different levels of agreement or impact on each factor.
Based on the density of circles, "Reduced transaction costs" and "Enabled access for a new generation of investors" appear to have received more positive responses than the other two factors.3. Understanding Staking:
The panel presents four circles with yellow dots, each representing a different response to the question of understanding staking.
The responses are:
• "Yes, I understand it well and see its potential benefits"
• "Yes, I understand it, but I am skeptical about its benefits"
• "I have a basic understanding but would like more detailed information"
• "No, I am not familiar with staking at all"Based on the density of dots, the most common response seems to be "Yes, I understand it well and see its potential benefits", followed by "Yes, I understand it, but I am skeptical about its benefits". This suggests that a significant portion of the respondents have some knowledge of staking but might have varying levels of confidence or understanding.
In my opinion...
I dont know enough to comment about Cryptocurrency/Blockchain, the data suggested that a moderate majority of respondents viewed Bitcoin as moderately viable as an alternative currency.
Attendeee Map
Original Image
Model Output
Text
Title:
World Map Pins
Description:
The image shows a world map with a purple background displayed on a black surface. The map has numerous small, yellow-orange dots scattered across the continents, indicating various locations. The question "WHERE IN THE WORLD ARE YOU FROM?" is written in white bold letters at the top of the map.
Datapoints and Insights:
The map likely represents the origins of people or data points, with each yellow-orange dot representing a location. Based on the distribution of dots, it can be inferred that the data points have a global presence, with some areas having a higher concentration of dots than others.
• High Concentration: There is a higher concentration of dots in North America, Europe, and parts of Asia, suggesting a higher representation of individuals or data points from these regions.
• Lower Concentration: There are fewer dots in South America, Africa, and Australia, indicating a potentially lower representation of individuals or data points from these regions.Without further context, it is difficult to determine the exact meaning of the dots or the purpose of the map. However, the map and question suggest a possible focus on global origins or diversity.
In my opinion...
Creativity can't be forced, it has to be nurtured. For me, this happens best in environments that embrace diversity and inclusivity. Living in India, UAE, and Canada has exposed me to a wealth of perspectives, English is not my first language so collaborating with people from different backgrounds makes me feel empowered and constantly sparks new ideas. This wouldn't have been possible without the unwavering support of my family, and I'm deeply grateful for them.
Geimini Experimental Model
Learn more about available models here. Disclaimer by GCP:
The maximum lifespan for a custom model is 18 months as of the GA release. You must create and train a new model to continue classifying content after that amount of time.
Lessons Learnt
- Technical skill- hands-on with prompt engineering, and AI-powered data analysis (AI-BI use-cases?).
- Impact of Automation, AI, Earning Public's Trust, and Diverse Perspectives
Ultimately, I'm incredibly grateful for the opportunity to attend Collision Conf and learn from such a fantastic community. As a data enthusiast, I'm passionate about exploring the intersection of AI and various industries.
Feel free to connect and share your thoughts on the future of user feedback analysis!
Top comments (0)